feat: add client availability tracking and error handling improvements

- Introduced `IsAvailable` and `SetUnavailable` methods to clients for availability tracking.
- Integrated availability checks in client selection logic to skip unavailable clients.
- Enhanced error handling by marking clients unavailable on specific error codes (e.g., 401, 402).
- Removed redundant quota verification logs in client reordering logic.

internal/client/claude_client.go

@@ -67,6 +67,7 @@ func NewClaudeClient(cfg *config.Config, ts *claude.ClaudeTokenStorage) *ClaudeC
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
 		claudeAuth:  claude.NewClaudeAuth(cfg),
 		apiKeyIndex: -1,
@@ -102,6 +103,7 @@ func NewClaudeClientWithKey(cfg *config.Config, apiKeyIndex int) *ClaudeClient {
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       &empty.EmptyStorage{},
+			isAvailable:        true,
 		},
 		claudeAuth:  claude.NewClaudeAuth(cfg),
 		apiKeyIndex: apiKeyIndex,
@@ -581,3 +583,13 @@ func (c *ClaudeClient) IsModelQuotaExceeded(model string) bool {
 func (c *ClaudeClient) GetRequestMutex() *sync.Mutex {
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *ClaudeClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *ClaudeClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/client.go

@@ -41,6 +41,9 @@ type ClientBase struct {
 
 	// modelRegistry is the global model registry for tracking model availability.
 	modelRegistry *registry.ModelRegistry
+
+	// unavailable tracks whether the client is unavailable
+	isAvailable bool
 }
 
 // GetRequestMutex returns the mutex used to synchronize requests for this client.

internal/client/codex_client.go

@@ -65,6 +65,7 @@ func NewCodexClient(cfg *config.Config, ts *codex.CodexTokenStorage) (*CodexClie
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
 		codexAuth:   codex.NewCodexAuth(cfg),
 		apiKeyIndex: -1,
@@ -100,6 +101,7 @@ func NewCodexClientWithKey(cfg *config.Config, apiKeyIndex int) *CodexClient {
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       &empty.EmptyStorage{},
+			isAvailable:        true,
 		},
 		codexAuth:   codex.NewCodexAuth(cfg),
 		apiKeyIndex: apiKeyIndex,
@@ -557,3 +559,13 @@ func (c *CodexClient) IsModelQuotaExceeded(model string) bool {
 func (c *CodexClient) GetRequestMutex() *sync.Mutex {
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *CodexClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *CodexClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/gemini-cli_client.go

@@ -69,6 +69,7 @@ func NewGeminiCLIClient(httpClient *http.Client, ts *geminiAuth.GeminiTokenStora
 			cfg:                cfg,
 			tokenStorage:       ts,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 	}
 
@@ -871,7 +872,18 @@ func (c *GeminiCLIClient) GetRequestMutex() *sync.Mutex {
 	return nil
 }
 
+// RefreshTokens is not applicable for Gemini CLI clients as they use API keys.
 func (c *GeminiCLIClient) RefreshTokens(ctx context.Context) error {
 	// API keys don't need refreshing
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *GeminiCLIClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *GeminiCLIClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/gemini-web_client.go

@@ -96,6 +96,7 @@ func NewGeminiWebClient(cfg *config.Config, ts *gemini.GeminiWebTokenStorage, to
 			cfg:                cfg,
 			tokenStorage:       ts,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 		tokenFilePath:  tokenFilePath,
 		convStore:      make(map[string][]string),
@@ -1072,3 +1073,13 @@ func (c *GeminiWebClient) storeConversationJSON(model string, history []geminiWe
 	c.convMutex.Unlock()
 	_ = geminiWeb.SaveConvData(geminiWeb.ConvDataPath(c.tokenFilePath), items, index)
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *GeminiWebClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *GeminiWebClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/gemini_client.go

@@ -54,6 +54,7 @@ func NewGeminiClient(httpClient *http.Client, cfg *config.Config, glAPIKey strin
 			httpClient:         httpClient,
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 		glAPIKey: glAPIKey,
 	}
@@ -445,3 +446,13 @@ func (c *GeminiClient) RefreshTokens(ctx context.Context) error {
 	// API keys don't need refreshing
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *GeminiClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *GeminiClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/openai-compatibility_client.go

@@ -68,6 +68,7 @@ func NewOpenAICompatibilityClient(cfg *config.Config, compatConfig *config.OpenA
 			httpClient:         httpClient,
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
+			isAvailable:        true,
 		},
 		compatConfig:       compatConfig,
 		currentAPIKeyIndex: apiKeyIndex,
@@ -425,3 +426,13 @@ func (c *OpenAICompatibilityClient) RefreshTokens(ctx context.Context) error {
 func (c *OpenAICompatibilityClient) GetRequestMutex() *sync.Mutex {
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *OpenAICompatibilityClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *OpenAICompatibilityClient) SetUnavailable() {
+	c.isAvailable = false
+}

internal/client/qwen_client.go

@@ -61,6 +61,7 @@ func NewQwenClient(cfg *config.Config, ts *qwen.QwenTokenStorage) *QwenClient {
 			cfg:                cfg,
 			modelQuotaExceeded: make(map[string]*time.Time),
 			tokenStorage:       ts,
+			isAvailable:        true,
 		},
 		qwenAuth: qwen.NewQwenAuth(cfg),
 	}
@@ -447,3 +448,13 @@ func (c *QwenClient) IsModelQuotaExceeded(model string) bool {
 func (c *QwenClient) GetRequestMutex() *sync.Mutex {
 	return nil
 }
+
+// IsAvailable returns true if the client is available for use.
+func (c *QwenClient) IsAvailable() bool {
+	return c.isAvailable
+}
+
+// SetUnavailable sets the client to unavailable.
+func (c *QwenClient) SetUnavailable() {
+	c.isAvailable = false
+}